Re-fitting improper dihedral parameters

Background

Improper parameters were re-fit for the first time in the Sage 2.1 release. However, there has still not been a whole lot of attention given to them. Moreover, we know that our functional form is not representative of certain groups, e.g. trivalent nitrogen, where the improper angle has two minima (https://openforcefieldgroup.slack.com/archives/C8P54J2JY/p1691613182717739 ).

To improve improper parameters, we could look at:

• changing up the functional form

• fitting to different data or targets

• making parameters more specific, should they need it – we have relatively few improper torsions